Главная > Продукты > Облачные вычисления и дата-центры  > HPC > Details on high performance computing

Inspur Yitian NF 5288M4

Inspur Yitian NF5288M4(Low-space high-performance isomeric server product for the client

Functional Features

Functional Features

Target customers and Life science:

molecular dynamics, gene sequencing, protein folding, computational chemistry, etc.

Engineering science: CAD/CAM/CAE, astrophysics, CFD, Mathematics and LabView

Government civil defense: weapon, image processing, battlefield simulation

Medicine: MRI, CT, image-aided treatment

Petroleum and petrochemical: seismic data processing, oil reservoir simulation

Finance: risk analysis, derivative financial product simulation, financial model establishment

Visualization: render farm, game, animation

Electronic design automation EDA: SPICE, Verilog, 3D EM

Product advantages

CPU+GPU isomeric computing architecture

Innovative introduction of GPU computing unit, which is a breakthrough in the conventional singular computing unit; latest processor core of Intel is adopted to work with NVIDIA Tesla/Kepler accelerated computing technology so that CPU and GPU work separately to perform their duties. CPU mainly performs logic selection, judgment skipping and IO communication in which it is an expert while GPU/MIC specializes in the intensive and highly parallel computing work, hence guaranteeing reasonable distribution of computing resources, fully releasing computing capacity and increasing the computing performance by several to several hundred folds. It is particularly noteworthy that concerted computing acceleration architecture does deny conventional technical routing but supplements and aids because it can give better rein to the advantages of GPU/MIC and finally help the user to substantially shorten the time of problem settlement or increase the scale of problem solved in specific applications.

Virtual solution of excellent performance

Compared with a conventional GPU virtualization solution, the Nvidia Kepler-based GRID K1 and K2 cards are specially designed to demonstrate more diversified effects in a virtualized environment. Meanwhile, it realizes the virtualization of GPU hardware, which means that multiple users can share the same GPU, hence improving user density and providing the real PC performances and compatibility.

Grid card can support the GPU-based virtualization solution of Citrix, Microsoft and VMware so that people can flexibly select from the extensive array of service-proven solutions.

Large-scale parallel computing and processing core

Compared with the multi-core CPU that allows very few threads to work simultaneously, GPU can implement several thousand threads at the same time, which allows our system to process more information flows. In game and animation rendering, for instance, presence of several million threads will allow simultaneous rendering of all pixels and the computing of all pixels can be completed within the time of one pixel because GPU computing is a special computing task, i.e. there is no need to consider the sequence of pixels to be computed.

Coordinate to settle inter-unit high-speed IO exchange

Each device has its special connection. GPU does not need to request for bandwidth from the bus and the data transmission rate can be increased substantially. Compared with a conventional PCI bus which only realizes unidirectional transmission within a single period of time, the two-way simplex connection of PCI-E can provide higher transmission rate and quality. Their difference is similar to that of half duplex and full duplex. Inspur “Yitian” NF5588M3 is furnished with PCI-E 3.0 x16, with a bandwidth of 32GB/s.

Latest QPI technology of Intel is adopted

In order to give better rein to the synergy effects of CPU and GPU, Inspur “Yitian” desktop supercomputer is furnished with Intel’s QPI (Quick Path Interconnect) technology so that a maximum transmission speed of 8.0GT/s can be realized to ensure faster communication. QPI bus allows internal interconnection of multi-core processor. In case of multiple processors, data can be transmitted between processors without the need for chip set, hence substantially increasing the general system performances. When a Sandybridge processor is integrated with memory controller, PCI-E 3.0 graphic interface and even image core, the advantages of QPI architecture will be further displayed.

Outstanding programming environment

CUDA™ (Compute Unified Device Architecture) generic parallel computing architecture enables GPU to solve a complicated computing problem. It comprises CUDA instruction set architecture (ISA) and the internal parallel computing engine in GPU. A developer can now use C language to program for CUDA™ architecture. C language is the most extensively-used advanced programming language. The programs compiled in this language can run in high performances on a processor that supports CUDA™.

Standard C language for GPU parallel application development

Standard database of Fast Fourier Transform Algorithm (FFT) and basic linear algebra subprogram (BLAS)

Special CUDA driver for fast data transmission and computing between GPU and CPU

CUDA drive program and OpenGL and DirectX graphic drive program can be inter-operated.

Support Linux 32/64 bits, Windows XP32/64 bits and Mac operating system

Technical Specifications

Inspur Yitian NF5288M4

Processor

Intel® Xeon® E5-2600V2 processor series up to 150W

Memory

Support 512GB DDR3 1600/1333/1066 RECC memory at the maximum; four-channel reading; support advanced memory error correction; memory mirroring; memory hot backup and other advanced features

SATA controller

Integrated with SATA disk controller; optional 8 channels SAS 3Gb and 6Gb disk controller

Memory

Support hot-plug disk capacity of 32TB at the maximum

PCI expansion slot

Support 4 Nvidia GPU card or Intel MIC card at the maximum

I/O port

Integrated with serial port, RJ45 network interface, USB2.0 port, independent IPMI management network interface and VGA interface

Power source

1+1 backup 1620W 80PLUS Platinum power source

Operating system

Microsoft Windows Server 2008 SP1 32/64bit

Windows 2003 Enterprise with SP2 32/64bit

Red Hat Enterprise Linux 5U3 32/64bit

SuSE Linux Enterprise Server 10 SP2 32/64bit

SuSE Linux Enterprise Server 11 32/64Bit

The OS has different supporting ability in different configurations. Consult with Inspur for more details.

Other information

Can support Nvidia Tesla K20/K20X

Can support Nvidia Grid K1/K2

User Manual

 
Related Products

中科汇联承办,easysite内容管理系统,portal门户,舆情监测,搜索引擎,政府门户,信息公开,电子政务